Thursday, January 19, 2006

Gonzales v Google: the study methodology is flawed

The text below was sent to Joel McElvain at the Department of Justice:

I have read and considered your motion to force Google to comply. May I humbly suggest that you may win the battle and lose the war. I have no stake in this dispute, but I do know a few things about computer technology. Yes, you will find me with a Google search.

Assuming you get the data requested from Google and its competitors, what kind of picture can you paint with the data obtained for the Supreme Court? The answer is not enough that is useful in setting United States law and policy. Internet search providers such as Google have search servers spread throughout the world, largely to handle queries in local geographies. The search engines must comply with local laws such as Arabic bans on indecency, China's ban on "democracy" and France's ban on Nazi memorabilia auctions.

For various reasons, Internet data that is searchable and available in one part of the world may not be available in or accessible from the United States, and vice versa. Therefore, it appears to me that a random set of data from Google's query servers around the world cannot say whether the query made in, say, China can also return the same results if made from the United States -- and therefore be controlled by U.S. law. Your subpoena will drag in worldwide data without identifying the query limits of the source data server.

As a citizen, I am interested in seeing that the Supreme Court's remand re COPA is met with a study that is accurate and defensible as it relates to U.S. law and citizens. I fail to see how the methodology implied in the Google motion is going to achieve a "national" set of data when a "world" net is being cast.

Peter S. Kastner

No comments:

Post a Comment

All comments are moderated.

Note: Only a member of this blog may post a comment.